update

app.py

@@ -21,7 +21,7 @@ from transformers import LlamaTokenizer
 
 print(datetime.datetime.now())
 
-NUM_THREADS = 4
+NUM_THREADS = 1
 
 print(NUM_THREADS)
 
@@ -166,8 +166,27 @@ def evaluate(
     top_k: int = 0
     prompt = generate_prompt(instruction, input)
     inputs = tokenizer(prompt, return_tensors="pt")
-    if len(inputs["input_ids"][0]) > max_tokens:
-        return f"please reduce length. Currently, {len(inputs['input_ids'][0])} token are used.", gr.update(interactive=True), gr.update(interactive=True)
+    if len(inputs["input_ids"][0]) > max_tokens + 10:
+        if HF_TOKEN and DATASET_REPOSITORY:
+            try:
+                now = datetime.datetime.now()
+                current_time = now.strftime("%Y-%m-%d %H:%M:%S")
+                print(f"[{current_time}] Pushing prompt and completion to the Hub")
+                save_inputs_and_outputs(
+                    now,
+                    prompt,
+                    "",
+                    {
+                        "temperature": temperature,
+                        "top_p": top_p,
+                        "top_k": top_k,
+                        "num_beams": num_beams,
+                        "max_tokens": max_tokens,
+                    },
+                )
+            except Exception as e:
+                print(e)
+        return f"please reduce the input length. Currently, {len(inputs['input_ids'][0])} token are used.", gr.update(interactive=True), gr.update(interactive=True)
     input_ids = inputs["input_ids"].to(device)
     generation_config = GenerationConfig(
         temperature=temperature,